This research seeks to develop a quantitative model of the development of habitual or skilled performance which can predict the cumulative nature and stability of learning and human information processing. Human information processing can be categorized as the result of two qualitatively different processes, controlled and automatic information processing. Control processing requires little training to initiate, is easy to modify, but slow, serial in nature, and resource-consumptive. Automatic processing occurs when a process is well-learned and automatized in long-term memory after consistent training. Automatic processes are fast, parallel, require little or no attentional resources to operate, but are difficult to modify. This research applies automatic/control processing concepts to heterarchy information processing. Heterarchal processes are ones which build complex information processing systems by incorporating modular, semi-autonomous components into multiple processes. This research will determine: 1) the conditions necessary to develop an automatic process heterarchy; 2) show how lower level automatic units are combined into higher level units; 3) training characteristics for optimal development of automatic heterarchies; 4) resource costs of execution of these heterarchies; 5) influence of context on processing output; 6) difficulty of modification of heterarchies; 7) resource costs of operating at multiple levels; and 8) the effects of slowing the sequential automatic processing. In addition to explicating the characteristics of human habit, skill development, and attention, this work may have important implications for context control of behavior, behavior modification, and sustained attention.